Re: [BUG 2.6.32] md/raid1: barrier disabling does not work correctlyin all cases

On Thu, Jan 20, 2011 at 3:25 PM, NeilBrown wrote:

> Normally the first thing that md/raid1 writes to a member device is t=
he
> metadata. =A0This is written with a barrier write if possible. =A0If =
that fails
> then barriers_work is cleared, so all barrier writes from the filesys=
tem
> (empty or otherwise) will be rejected.

Yes, I noticed that, and assume that's why more people haven't seen
this problem...

> As you are getting an error here I assume that you are using non-pers=
istent
> metadata - correct?

Yep, that's right...

> Nonetheless, I think the correct fix is to add a special case for ret=
rying a
> zero-length.

> Something like this maybe?

Thanks Neil. I will try this patch.

--
Paul
--
To unsubscribe from this list: send the line "unsubscribe linux-raid" i=
n
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html